Use sqlite .coverage file to do cool stuff

[spookylukey]

This is a PR that combines a lot of stuff (mentioned in #6), so I'm not expecting it to be merged as is, but I thought I'd share at this point even if it's not quite ready. I have been using this for several months now and found it useful.

Also my elisp is probably pretty bad!

Basically 2 main features:

1. The ability to use the .coverage SQLite file directly, without needing the coverage.xml report to be generated, for normal overlay analysis. The SQLite file is less preferred, because the XML file contains processing done to handling empty lines and docstrings etc,. - analysis that could not easily be reproduced in elisp, it requires a Python parser.

   This requires using some internals of the coverage SQLite schema, and reproducing some of the "packing" logic for line numbers (see https://github.com/nedbat/coveragepy/blob/master/coverage/numbits.py ). However, we include version checking of the file, and if the schema changes in the future, we can add support for multiple schemas easily.

2. If --cov-context=test has been used, it adds the ability to find which tests executed which lines of code. This can only be implemented using the SQLite file, as the XML doesn't contain this info. The only user interface on this at the moment is the command python-coverage-rerun-pytest-current-region command, which is the only interface I needed. It basically allows you to put your cursor somewhere in the code, do M-x python-coverage-rerun-pytest-current-region and get a pytest command line for rerunning the tests that executed that code, like this:

pytest path/to/tests.py::test_1 path/to/tests.py::test_2 

This is super useful on some occasions!